Subject: Re: pageinspect: Hash index support
To: Tom Lane <tgl@sss.pgh.pa.us>
References: <e2ac6c58-b93f-9dd9-f4e6-d6d30add7fdf@redhat.com>
 <e1bda3f7-dc99-c610-4d48-58815f5b7515@redhat.com>
 <CAE9k0P=OQSEj0x9z91tvE9P1f-Ht7x7z8X+eDWQGTJNSFbn9yg@mail.gmail.com>
 <5d64a536-cc90-ce96-b68f-5b6bd3206476@redhat.com>
 <CAB7nPqTT3ZEm8QKsYYxfUnWL+P3p2mAbud7AFiZpK_JKx3eADA@mail.gmail.com>
 <d235dcd6-146a-1269-73a0-da9a55bfeddc@redhat.com>
 <80a00b5c-56d3-dffb-8dba-edf5d2167b12@2ndquadrant.com>
 <16d89e7e-cb1a-cb5d-e944-6f10c324a312@redhat.com>
 <c982b01c-06bc-485b-5dde-057ee244261a@2ndquadrant.com>
 <836487ee-8d4e-cd49-54ac-f4793ddba5ae@redhat.com>
 <8a751498-6ee9-b943-abf3-b5641851d7de@2ndquadrant.com>
 <62ee2eba-3a5b-06d3-e224-cfe00b395be7@redhat.com>
 <f52f2117-627b-baa3-1e83-59394e881bf5@2ndquadrant.com>
 <f8a19652-c5e4-4b8f-a461-62d886ff2400@2ndquadrant.com>
 <66afc31d-e97a-41c3-da4c-bfa02fa2d54e@redhat.com>
 <eb68fb02-5a85-e643-11cf-9319dd7edf2b@2ndquadrant.com>
 <02dc6cef-a80c-c976-8ac0-518ff9b22dd8@2n! dquadrant.com>
 <8921.1478109287@sss.pgh.pa.us>
Cc: Jesper Pedersen <jesper.pedersen@redhat.com>,
 Michael Paquier <michael.paquier@gmail.com>,
 Ashutosh Sharma <ashu.coek88@gmail.com>, Jeff Janes <jeff.janes@gmail.com>,
 Alvaro Herrera <alvherre@2ndquadrant.com>,
 PostgreSQL Hackers <pgsql-hackers@postgresql.org>
From: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Organization: 2ndQuadrant
Message-ID: <7351c7bb-db87-be4a-0b86-7f7aad0bed41@2ndquadrant.com>
Date: Thu, 3 Nov 2016 11:28:33 -0400
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:45.0)
 Gecko/20100101 Thunderbird/45.4.0
MIME-Version: 1.0
In-Reply-To: <8921.1478109287@sss.pgh.pa.us>
Content-Type: multipart/mixed;
 boundary="------------49C01783DFF74C1295E7DFA2"
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

This is a multi-part message in MIME format.
--------------49C01783DFF74C1295E7DFA2
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit

On 11/2/16 1:54 PM, Tom Lane wrote:
> I think the right thing is likely to be to copy the presented bytea
> into a palloc'd (and therefore properly aligned) buffer.  And not
> just in this one function.

Does the attached look reasonable?

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

--------------49C01783DFF74C1295E7DFA2
Content-Type: text/x-patch;
 name="0001-pageinspect-Fix-unaligned-struct-access-in-GIN-funct.patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename*0="0001-pageinspect-Fix-unaligned-struct-access-in-GIN-funct.pa";
 filename*1="tch"

From 086c603016de5018a8baddc88a8932472e0fcd1e Mon Sep 17 00:00:00 2001
From: Peter Eisentraut <peter_e@gmx.net>
Date: Thu, 3 Nov 2016 12:00:00 -0400
Subject: [PATCH] pageinspect: Fix unaligned struct access in GIN functions

The raw page data that is passed into the functions will not be aligned
at 8-byte boundaries.  Casting that to a struct and accessing int64
fields will result in unaligned access.  On most platforms, you get away
with it, but it will result on a crash on pickier platforms such as ia64
and sparc64.
---
 contrib/pageinspect/ginfuncs.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/contrib/pageinspect/ginfuncs.c b/contrib/pageinspect/ginfuncs.c
index 145e2ce..660d236 100644
--- a/contrib/pageinspect/ginfuncs.c
+++ b/contrib/pageinspect/ginfuncs.c
@@ -51,7 +51,10 @@ gin_metapage_info(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 				 errmsg("input page too small (%d bytes)", raw_page_size)));
-	page = VARDATA(raw_page);
+
+	/* make a copy so that the page is properly aligned for struct access */
+	page = palloc(raw_page_size);
+	memcpy(page, VARDATA(raw_page), raw_page_size);
 
 	opaq = (GinPageOpaque) PageGetSpecialPointer(page);
 	if (opaq->flags != GIN_META)
@@ -115,7 +118,10 @@ gin_page_opaque_info(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 				 errmsg("input page too small (%d bytes)", raw_page_size)));
-	page = VARDATA(raw_page);
+
+	/* make a copy so that the page is properly aligned for struct access */
+	page = palloc(raw_page_size);
+	memcpy(page, VARDATA(raw_page), raw_page_size);
 
 	opaq = (GinPageOpaque) PageGetSpecialPointer(page);
 
@@ -195,7 +201,10 @@ gin_leafpage_items(PG_FUNCTION_ARGS)
 			ereport(ERROR,
 					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 				  errmsg("input page too small (%d bytes)", raw_page_size)));
-		page = VARDATA(raw_page);
+
+		/* make a copy so that the page is properly aligned for struct access */
+		page = palloc(raw_page_size);
+		memcpy(page, VARDATA(raw_page), raw_page_size);
 
 		if (PageGetSpecialSize(page) != MAXALIGN(sizeof(GinPageOpaqueData)))
 			ereport(ERROR,
-- 
2.10.2


--------------49C01783DFF74C1295E7DFA2
Content-Type: text/plain
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
MIME-Version: 1.0


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

--------------49C01783DFF74C1295E7DFA2--